IFND: a benchmark dataset for fake news detection.

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Science

Background:

• Fake news propagation on social media causes confusion and biased perspectives.
• Existing fake news identification methods face challenges due to a lack of comprehensive, publicly available benchmark datasets, especially for Indian news.
• This paper introduces the Indian Fake News Dataset (IFND), a large-scale, multi-modal dataset containing both text and images of Indian news from 2013-2021.

Discussion:

• The IFND dataset was created by scraping content and enhanced using an intelligent augmentation algorithm for generating meaningful fake news statements.
• Topic modeling using Latent Dirichlet Allocation (LDA) was employed to categorize news statements.
• Machine learning, deep learning, and multi-modal approaches (textual and visual features) were implemented to evaluate the dataset's performance.

Key Insights:

• The proposed IFND dataset achieved satisfactory results in fake news detection tasks.
• The dataset's multi-modal nature (text and image) allows for robust analysis.
• The creation and release of IFND address the critical need for a large-scale, Indian-centric benchmark dataset.

Outlook:

• The availability of the IFND dataset is expected to stimulate further research in fake news detection.
• This resource can lead to the development of more accurate and effective prediction models for combating misinformation.
• Future work may involve expanding the dataset and exploring advanced multi-modal fusion techniques.